Programming Languages and Translators - … › ~sedwards › classes › 2003 › w4115f ›...
Transcript of Programming Languages and Translators - … › ~sedwards › classes › 2003 › w4115f ›...
Programming Languages andTranslatorsCOMS W4115
Pieter Bruegel, The Tower of Babel, 1563
Prof. Stephen A. EdwardsFall 2003
Columbia UniversityDepartment of Computer Science
Instructor
Prof. Stephen A. [email protected]://www.cs.columbia.edu/˜sedwards/462 Computer Science BuildingOffice Hours: 4–5 PM Tuesday, Thursday
Schedule
Tuesdays and Thursdays, 5:40 PM to 6:55 PM
Room 717, Hamilton Hall
September 2 to December 4
Midterm: October 14
Holidays: November 4 (Election day), November 27(Thanksgiving)
Objectives
Theory of language design
• Finer points of languages
• Different languages and paradigms
Practice of Compiler Construction
• Overall structure of a compiler
• Automated tools and their use
• Lexical analysis to assembly generation
Required Text
Alfred V. Aho, Ravi Sethi, andJeffrey D. Ullman.Compilers: Principles, Techniques,and Tools.Addison-Wesley, 1985.
Available from Papyrus, 114th andBroadway.
Assignments and Grading
40% Programming Project
20% Midterm (near middle of term)
30% Final (at end of term)
10% Individual homework
Bottom line: do well on the project, you’ll get a good grade.
Prerequisite: COMS W3156Software Engineering
Teams will build a large software system
Makefiles, version control, test suites
Testing will be as important as development
Prerequisite:COMS W3261 Computability
You need to understand grammars.
We will be working with regular and context-freelanguages.
Class Website
Off my home page,http://www.cs.columbia.edu/˜sedwards/
Contains syllabus, lecture notes, and assignments.
Schedule will be continually updated during the semester.
Collaboration
Collaborate with your team on the project.
Homework is to be done by yourself.
Tests: Will be closed book.
The Project
The Project
Design and implement your own little language.
Five deliverables:
1. A white paper describing and motivating yourlanguage
2. A language reference manual defining it formally
3. A compiler or interpreter for your language running onsome sample programs
4. A final project report
5. A final project presentation
Teams
Immediately start forming four-person teams to work onthis project.
Each team will develop its own langauge.
Suggested division of labor: Front-end, back-end, testing,documentation.
All members of the team should be familiar with the wholeproject.
First Three Tasks
1. Decide who you will work with
You’ll be stuck with them for the term; choose wisely.
2. Elect a team leader
Languages come out better from dictatorships, notdemocracies. Besides, you’ll have someone to blame.
3. Select a weekly meeting time
Harder than you might think. Might want to discusswith a TA you’d like to have so it is convenient forhim/her as well.
White Paper
Follow the style of the Java white paper (see the classwebsite for a link).
4–8 pages.
Answer the question, “why another language?” with adescription of what your language is intended for.
Small snippets of code to show syntax is enough.
Language Reference Manual
A careful definition of the syntax and semantics of yourlanguage.
Follow the style of the C language reference manual(Appendix A of Kernighan and Ritchie, The CProgramming Langauge; see the class website).
Final Report Sections
1. Introduction: the white paper
2. Language Tutorial
3. Language Reference Manual
4. Project Plan
5. Architectural Design
6. Test Plan
7. Lessons Learned
8. Complete listing
Due Dates
White Paper September 23 soon
Reference Manual October 23
Final Report December 12?
Design a language?
A small, domain-specific language.
Think of awk or php, not Java or C++.
Examples from last term:
Quantum computing language
Geometric figure drawing language
Projectile motion simulation langauge
Matlab-like array manipulation language
Screenplay animation language
Other language ideas
Simple animation language
Model train simulation language
Escher-like pattern generator
Music manipulation language (harmony)
Web surfing language
Mathematical function manipulator
Simple scripting language (a la Tcl)
Petri net simulation language
What’s in aLanguage?
Components of a language: Syntax
How characters combine to form words, sentences,paragraphs.
The quick brown fox jumps over the lazy dog.
is syntactically correct English, but isn’t a Java program.
class Foo {
public int j;
public int foo(int k) { return j + k; }
}
Is syntactically correct Java, but isn’t C.
Specifying Syntax
Usually done with a context-free grammar.
Typical syntax for algebraic expressions:
expr → expr + expr
| expr − expr
| expr ∗ expr
| expr/expr
| digit
| (expr)
Components of a language:Semantics
What a well-formed program “means.”
The semantics of C says this computes the nth Fibonaccinumber.
int fib(int n){
int a = 0, b = 1;int i;for (i = 1 ; i < n ; i++) {
int c = a + b;a = b;b = c;
}return b;
}
Semantics
Something may be syntactically correct but semanticallynonsensical.
The rock jumped through the hairy planet.
Or ambiguous
The chickens are ready for eating.
Semantics
Nonsensical in Java:
class Foo {
int bar(int x) { return Foo; }
}
Ambiguous in Java:
class Bar {
public float foo() { return 0; }
public int foo() { return 0; }
}
Specifying Semantics
Doing it formally beyond the scope of this class, butbasically two ways:
• Operational semantics
Define a virtual machine and how executing theprogram evolves the state of the virtual machine
• Denotational semantics
Shows how to build the function representing thebehavior of the program (i.e., a transformation ofinputs to outputs) from statements in the language.
Most language definitions use an informal operationalsemantics written in English.
Great Moments inProgramming Language Evolution
Assembly
Before: numbers5589E58B45088B550C39D0740D39D07E0829D039D075F6C9C329C2EBF6
After: Symbolsgcd: pushl %ebp
movl %esp, %ebpmovl 8(%ebp), %eaxmovl 12(%ebp), %edxcmpl %edx, %eaxje .L9
.L7: cmpl %edx, %eaxjle .L5subl %edx, %eax
.L2: cmpl %edx, %eaxjne .L7
.L9: leaveret
.L5: subl %eax, %edxjmp .L2
FORTRAN
Beforegcd: pushl %ebp
movl %esp, %ebpmovl 8(%ebp), %eaxmovl 12(%ebp), %edxcmpl %edx, %eaxje .L9
.L7: cmpl %edx, %eaxjle .L5subl %edx, %eax
.L2: cmpl %edx, %eaxjne .L7
.L9: leaveret
.L5: subl %eax, %edxjmp .L2
After: Expressions, control-flow10 if (a .EQ. b) goto 20
if (a .LT. b) thena = a - b
elseb = b - a
endifgoto 10
20 end
COBOL
Added type declarations, record types, file manipulation
data division.file section.* describe the input filefd employee-file-in
label records standardblock contains 5 recordsrecord contains 31 charactersdata record is employee-record-in.
01 employee-record-in.02 employee-name-in pic x(20).02 employee-rate-in pic 9(3)v99.02 employee-hours-in pic 9(3)v99.02 line-feed-in pic x(1).
LISP, Scheme, Common LISP
Functional, high-level languages
(defun gnome-doc-insert ()"Add a documentation header to the current function.
Only C/C++ function types are properly supported currently."(interactive)(let (c-insert-here (point))(save-excursion
(beginning-of-defun)(let (c-arglist
c-funcname(c-point (point))c-comment-pointc-isvoidc-doinsert)
(search-backward "(")(forward-line -2)(while (or (looking-at "ˆ$")
(looking-at "ˆ *}")(looking-at "ˆ \\*")(looking-at "ˆ#"))
(forward-line 1))
APL
Powerful operators, interactive language
Source: Jim Weigang, http://www.chilton.com/˜jimw/gsrand.html
Algol, Pascal, Clu, Modula, Ada
Imperative, block-structured language, formal syntaxdefinition, structured programmingPROC insert = (INT e, REF TREE t)VOID:
# NB inserts in t as a side effect #IF TREE(t) IS NIL THEN t := HEAP NODE := (e, TREE(NIL), TREE(NIL))ELIF e < e OF t THEN insert(e, l OF t)ELIF e > e OF t THEN insert(e, r OF t)FI;
PROC trav = (INT switch, TREE t, SCANNER continue, alternative)VOID:# traverse the root node and right sub-tree of t only. #IF t IS NIL THEN continue(switch, alternative)ELIF e OF t <= switch THEN
print(e OF t);traverse( switch, r OF t, continue, alternative)
ELSE # e OF t > switch #PROC defer = (INT sw, SCANNER alt)VOID:
trav(sw, t, continue, alt);alternative(e OF t, defer)
FI;
Algol-68, source http://www.csse.monash.edu.au/˜lloyd/tildeProgLang/Algol68/treemerge.a68
SNOBOL, Icon
String-processing languagesLETTER = ’ABCDEFGHIJKLMNOPQRSTUVWXYZ$#@’SP.CH = "+-,=.*()’/& "SCOTA = SP.CHSCOTA ’&’ =Q = "’"QLIT = Q FENCE BREAK(Q) QELEM = QLIT | ’L’ Q | ANY(SCOTA) | BREAK(SCOTA) | REMF3 = ARBNO(ELEM FENCE)B = (SPAN(’ ’) | RPOS(0)) FENCEF1 = BREAK(’ ’) | REMF2 = F1CAOP = (’LCL’ | ’SET’) ANY(’ABC’) |
+ ’AIF’ | ’AGO’ | ’ACTR’ | ’ANOP’ATTR = ANY(’TLSIKN’)ELEMC = ’(’ FENCE *F3C ’)’ | ATTR Q | ELEMF3C = ARBNO(ELEMC FENCE)ASM360 = F1 . NAME B
+ ( CAOP . OPERATION B F3C . OPERAND |+ F2 . OPERATION B F3 . OPERAND)+ B REM . COMMENT
SNOBOL: Parse IBM 360 assembly. From Gimpel’s book, http://www.snobol4.org/
BASIC
Programming for the masses
10 PRINT "GUESS A NUMBER BETWEEN ONE AND TEN"
20 INPUT A$
30 IF A$ = "5" THEN PRINT "GOOD JOB, YOU GUESSED IT"
40 IF A$ = "5" GOTO 100
50 PRINT "YOU ARE WRONG. TRY AGAIN"
60 GOTO 10
100 END
Simula, Smalltalk, C++, Java, C#
The object-oriented philosophy
class Shape(x, y); integer x; integer y;virtual: procedure draw;begin
comment -- get the x & y components for the object --;integer procedure getX;
getX := x;integer procedure getY;
getY := y;
comment -- set the x & y coordinates for the object --;integer procedure setX(newx); integer newx;
x := newx;integer procedure setY(newy); integer newy;
y := newy;end Shape;
C
Efficiency for systems programming
int gcd(int a, int b)
{
while (a != b) {
if (a > b) a -= b;
else b -= a;
}
return a;
}
ML, Miranda, Haskell
Purer functional languagestructure RevStack = structtype ’a stack = ’a listexception Emptyval empty = []fun isEmpty (s:’a stack):bool =(case s
of [] => true| _ => false)
fun top (s:’a stack): =(case s
of [] => raise Empty| x::xs => x)
fun pop (s:’a stack):’a stack =(case s
of [] => raise Empty| x::xs => xs)
fun push (s:’a stack,x: ’a):’a stack = x::sfun rev (s:’a stack):’a stack = rev (s)
end
sh, awk, perl, tcl, python
Scripting languages:glue for binding the universe together
class() {classname=‘echo "$1" | sed -n ’1 s/ *:.*$//p’‘parent=‘echo "$1" | sed -n ’1 s/ˆ.*: *//p’‘hppbody=‘echo "$1" | sed -n ’2,$p’‘
forwarddefs="$forwarddefsclass $classname;"
if (echo $hppbody | grep -q "$classname()"); thendefaultconstructor=
elsedefaultconstructor="$classname() {}"
fi}
VisiCalc, Lotus 1-2-3, Excel
The spreadsheet style of programming
A B
1 Hours 23
2 Wage per hour $ 5.36
3
4 Total Pay = B1 * B2
SQL
Database queries
CREATE TABLE shirt (id SMALLINT UNSIGNED NOT NULL AUTO_INCREMENT,style ENUM(’t-shirt’, ’polo’, ’dress’) NOT NULL,color ENUM(’red’, ’blue’, ’white’, ’black’) NOT NULL,owner SMALLINT UNSIGNED NOT NULL
REFERENCES person(id),PRIMARY KEY (id)
);
INSERT INTO shirt VALUES(NULL, ’polo’, ’blue’, LAST_INSERT_ID()),(NULL, ’dress’, ’white’, LAST_INSERT_ID()),(NULL, ’t-shirt’, ’blue’, LAST_INSERT_ID());
Prolog
Logic Language
edge(a, b). edge(b, c).
edge(c, d). edge(d, e).
edge(b, e). edge(d, f).
path(X, X).
path(X, Y) :-
edge(X, Z), path(Z, Y).